Chuchu Fan
Chuchu Fan earned a bachelor’s degree at Tsinghua in 2013 and a doctorate at Illinois in 2019. She then held a Caltech postdoctoral position in 2019–2020. Her research group at MIT combines formal methods, machine learning, and control theory for autonomous systems. In 2025 she co-authored Def-MARL research on distributed multi-agent control with safety constraints. The paper uses an epigraph reformulation to improve training and reports simulator and quadcopter experiments. The results concern the tested scope and do not establish a universal safety guarantee.
- Works
- 161
- Citations
- 2,098
- h-index
- 25
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
2013–2020
Automation and systems research
Fan earned a Tsinghua bachelor’s degree in 2013 and an Illinois doctorate in 2019. She held a postdoctoral position at Caltech in 2019–2020.
-
As of October 2026
Learning with verification and control
Her research combines formal methods, control theory, and machine learning for the design and analysis of autonomous systems. Her work includes certified control and neural barrier certificates for several agents. These methods seek conditions under which system behaviour can be analysed.
-
2025
Distributed multi-agent learning
Fan co-authored an RSS 2025 paper on safe optimal control with distributed epigraph-form multi-agent reinforcement learning. The method, Def-MARL, reformulates constrained optimisation to improve training stability. It uses centralised training with distributed execution, allowing each agent to act within the resulting control framework.
-
2025
Tests on simulators and quadcopters
The paper reports experiments across two simulators and on Crazyflie quadcopters. The tests examine whether agents can complete collaborative tasks while satisfying safety constraints. The conclusions apply to those conditions and the method’s assumptions.
Sources and credits
Photo: aeroastro.mit.edu; creator not stated. License: Not stated by source. Source: aeroastro.mit.edu. Cropped to a square and resized. Ask us to remove or replace this photo.